Skip to content

Data / Dataset

Sections

Classes

Class Description
BlendedDatasetConfig Mixes multiple datasets together, sampling from each according to specified weights
ConcatenatedDatasetConfig Concatenate multiple indexed datasets as if they were one
DatasetConfig (abstract) Abstract base configuration for all dataset types
DatasetSliceConfig
IndexedDatasetConfig (abstract) Abstract configuration for indexed datasets that support random access by index
RedisConfig
SamplableDatasetConfig (abstract) Abstract configuration for datasets that can be built and then sampled
SampledDatasetConfig (abstract) A sampled dataset containing a prepared list of samples to be indexed sequentially (as-is) during training
SamplingConfig Holds all the necessary information for sampling
SamplingConfigBase A dataset-dependent configuration for sampling
StreamingDatasetConfig Configuration for a streaming dataset that reads training data from a Redis stream